Yup, two slots occupied with 256 MB each. It's non-interleaved, so it can be upgraded with single modules. I guess I'll go for a single 1GB module, and upgrade both boxes to 1GB. Half the fun with this vintage or "near vintage" boxes is in hunting for parts on auctions, I guess.
As to hostnames ... I can remember the good old times at the university when hosts had names like "gandalf, bilbo, kermit,misspiggy, spock, uhura ..." :-).
Those x330s have a service processor (actually a PowerPC) on-board which can be used to report all kind of things like temperatures, hardware failures etc. I haven't bothered yet to get this working under Linux, have you?
I guess I'll go for a single 1GB module, and upgrade both boxes to 1GB.
That will be a bit more expensive as single sticks of that size will command a premium. You should have no trouble (at auction) getting very cheap old machines with lots of 256MB sticks in them. You will be able to buy a whole machine with 4x256 cheaper than you can buy the single 1gig stick.
Quote:
Half the fun with this vintage or "near vintage" boxes is in hunting for parts on auctions, I guess.
Absolutely - very valuable therapy :).
Quote:
As to hostnames ... I can remember the good old times at the university when hosts had names like "gandalf, bilbo, kermit,misspiggy, spock, uhura ..." :-).
I liked 4 letter asteroid names so my first hosts were juno, nysa, leto, toro, zeus, hebe, etc, :).
Quote:
Those x330s have a service processor (actually a PowerPC) on-board which can be used to report all kind of things like temperatures, hardware failures etc. I haven't bothered yet to get this working under Linux, have you?
Nope!! That's well beyond my level of expertese :).
The long awaited memory for the 440 finally got here today! It's too bad half of it is dead. Yanked the top SMP module out, stuck the memory in, put the module back in, hooked up the SMP cables, and fired it up! Showed both modules, memory, a board light came on and it went into a reboot! Then I started trouble shooting the monster. Swapped everything around at least twice and then tried the new memory in another box, one module at a time. Two worked fine and two are dead. Naturally the 440 is 4-way interleave so you need four matching memory modules for it to work. Some days it just doesn't pay to get up in the morning! :(
This is the box now about 30% through the first 4 non-trashed results. Seems like each result will take just over 32 hours so the box will produce a result every 8 hours or so. This should be enough to make this box my best producer.
Not too shabby for someone else's junk :).
The machine I listed above has now completed 12 results (32hr average cruch time), and from the average of those 12, is headed for a RAC of around 700. At the moment its RAC is 0.0 since nobody else has returned any of the 12 results.
My previous best performer was a single cpu AMD64 3200+ overclocked a bit and having a RAC just below 500. So I'm very pleased with my essentially free acquisition :).
My p3-850celeron laptop just completed a 206 credit WU. Took 72hrs for a RAC of about 70.
Rather feeble, but I bought it because I wanted a ruggedized laptop to use for astronomical star charts and as a satnav system in my car (this feature pending).
The only really disappointing bit is that it's going to end up with a DCF of around 1.1, roughly twice as high as my a64x2 or c1d systems. Am I permanantly hobbled by ancient technology, or are there projects out there that are more old hardware friendly?
Finally, if there's anyone who's good at finding specific hardware in the results logs, I was wondering if the doubled cache from a non neutered p3 would make a significant difference. For a laptop the IX250 gobook is supposed to be rather easy to work on so a CPU upgrade isn't totally out of the question.
My p3-850celeron laptop just completed a 206 credit WU. Took 72hrs for a RAC of about 70.
This is a celeron 950 powered HP e-PC which is sort of like a ruggedised laptop without a screen. It is doing 232 credit results in times varying between 38 and 46 hours and it has a DCF of 0.64 and probably has a little more to drop yet.
Quote:
The only really disappointing bit is that it's going to end up with a DCF of around 1.1, roughly twice as high as my a64x2 or c1d systems. Am I permanantly hobbled by ancient technology, or are there projects out there that are more old hardware friendly?
1.1 for DCF is way too high for a coppermine celeron. On EAH your machine should beat a PIII 800 if it were running at full speed. I suspect it may not be running at 850MHz (ie maybe it's throttled down because of heat).
Quote:
Finally, if there's anyone who's good at finding specific hardware in the results logs, I was wondering if the doubled cache from a non neutered p3 would make a significant difference. For a laptop the IX250 gobook is supposed to be rather easy to work on so a CPU upgrade isn't totally out of the question.
Doesn't seem to make a difference. I've compared plenty of celeron 950s with PIII 933s and Celeron 1000 with PIII 1000 (both coppermine) and there's not much difference to see. I'd actually rather have the celeron as it's a lot easier to overclock by a reasonable amount, eg 100 -> 120 MHz.
This is a PIII 1.0G running at 1.15G and this is a celeron 1.0G running at 1.1G. When the crunch times vary the way they do it's hard to make accurate comparisons but my best guess is that the somewhat faster PIII is producing around 6.4 cr/hr and the somewhat slower celeron is producing around 6.3 cr/hr. Not a great deal of difference really.
1.1 for DCF is way too high for a coppermine celeron. On EAH your machine should beat a PIII 800 if it were running at full speed. I suspect it may not be running at 850MHz (ie maybe it's throttled down because of heat).
looks like you're right, whenever the fan's running it's throttled to 50%. I'm going to be out of town this weekend, but it looks like I'm going to have to crack the case and clean out the cooling. the fan itself is running rather eratically and but the actual progress has seemed smoother, I'll have to try messing with it more later.
Quote:
Doesn't seem to make a difference. I've compared plenty of celeron 950s with PIII 933s and Celeron 1000 with PIII 1000 (both coppermine) and there's not much difference to see. I'd actually rather have the celeron as it's a lot easier to overclock by a reasonable amount, eg 100 -> 120 MHz.
I will upgrade my two PIII-S boxes to 1GB each soon, I hope I've got more luck than Jim with his memory upgrade :-(. I will also do some power consumption tests then, hoping it won't be too shocking :-).
Hope you have better luck than I have with the memory up grade Bikeman. Seems strange, but, PC133r is more expensive than PC1600r or 2100r at present. Mine wouldn't be so bad except it's 4-way interleave and min size is 512MB modules. That, and IBM changes their FRU's like most people change their socks. Found several I think will work but need to cross reference the FRU's.
Yup, two slots occupied with
)
Yup, two slots occupied with 256 MB each. It's non-interleaved, so it can be upgraded with single modules. I guess I'll go for a single 1GB module, and upgrade both boxes to 1GB. Half the fun with this vintage or "near vintage" boxes is in hunting for parts on auctions, I guess.
As to hostnames ... I can remember the good old times at the university when hosts had names like "gandalf, bilbo, kermit,misspiggy, spock, uhura ..." :-).
Those x330s have a service processor (actually a PowerPC) on-board which can be used to report all kind of things like temperatures, hardware failures etc. I haven't bothered yet to get this working under Linux, have you?
CU
H-B
RE: I guess I'll go for a
)
That will be a bit more expensive as single sticks of that size will command a premium. You should have no trouble (at auction) getting very cheap old machines with lots of 256MB sticks in them. You will be able to buy a whole machine with 4x256 cheaper than you can buy the single 1gig stick.
Absolutely - very valuable therapy :).
I liked 4 letter asteroid names so my first hosts were juno, nysa, leto, toro, zeus, hebe, etc, :).
Nope!! That's well beyond my level of expertese :).
Cheers,
Gary.
The long awaited memory for
)
The long awaited memory for the 440 finally got here today! It's too bad half of it is dead. Yanked the top SMP module out, stuck the memory in, put the module back in, hooked up the SMP cables, and fired it up! Showed both modules, memory, a board light came on and it went into a reboot! Then I started trouble shooting the monster. Swapped everything around at least twice and then tried the new memory in another box, one module at a time. Two worked fine and two are dead. Naturally the 440 is 4-way interleave so you need four matching memory modules for it to work. Some days it just doesn't pay to get up in the morning! :(
RE: This is the box now
)
The machine I listed above has now completed 12 results (32hr average cruch time), and from the average of those 12, is headed for a RAC of around 700. At the moment its RAC is 0.0 since nobody else has returned any of the 12 results.
My previous best performer was a single cpu AMD64 3200+ overclocked a bit and having a RAC just below 500. So I'm very pleased with my essentially free acquisition :).
Cheers,
Gary.
My p3-850celeron laptop just
)
My p3-850celeron laptop just completed a 206 credit WU. Took 72hrs for a RAC of about 70.
Rather feeble, but I bought it because I wanted a ruggedized laptop to use for astronomical star charts and as a satnav system in my car (this feature pending).
The only really disappointing bit is that it's going to end up with a DCF of around 1.1, roughly twice as high as my a64x2 or c1d systems. Am I permanantly hobbled by ancient technology, or are there projects out there that are more old hardware friendly?
Finally, if there's anyone who's good at finding specific hardware in the results logs, I was wondering if the doubled cache from a non neutered p3 would make a significant difference. For a laptop the IX250 gobook is supposed to be rather easy to work on so a CPU upgrade isn't totally out of the question.
RE: My p3-850celeron laptop
)
This is a celeron 950 powered HP e-PC which is sort of like a ruggedised laptop without a screen. It is doing 232 credit results in times varying between 38 and 46 hours and it has a DCF of 0.64 and probably has a little more to drop yet.
1.1 for DCF is way too high for a coppermine celeron. On EAH your machine should beat a PIII 800 if it were running at full speed. I suspect it may not be running at 850MHz (ie maybe it's throttled down because of heat).
Doesn't seem to make a difference. I've compared plenty of celeron 950s with PIII 933s and Celeron 1000 with PIII 1000 (both coppermine) and there's not much difference to see. I'd actually rather have the celeron as it's a lot easier to overclock by a reasonable amount, eg 100 -> 120 MHz.
This is a PIII 1.0G running at 1.15G and this is a celeron 1.0G running at 1.1G. When the crunch times vary the way they do it's hard to make accurate comparisons but my best guess is that the somewhat faster PIII is producing around 6.4 cr/hr and the somewhat slower celeron is producing around 6.3 cr/hr. Not a great deal of difference really.
Cheers,
Gary.
RE: 1.1 for DCF is way too
)
looks like you're right, whenever the fan's running it's throttled to 50%. I'm going to be out of town this weekend, but it looks like I'm going to have to crack the case and clean out the cooling. the fan itself is running rather eratically and but the actual progress has seemed smoother, I'll have to try messing with it more later.
laptops generally aren't OCable though.
Hi all! I will upgrade my
)
Hi all!
I will upgrade my two PIII-S boxes to 1GB each soon, I hope I've got more luck than Jim with his memory upgrade :-(. I will also do some power consumption tests then, hoping it won't be too shocking :-).
CU
H-B
RE: I will also do some
)
Very punny... ;-)
Hope you have better luck
)
Hope you have better luck than I have with the memory up grade Bikeman. Seems strange, but, PC133r is more expensive than PC1600r or 2100r at present. Mine wouldn't be so bad except it's 4-way interleave and min size is 512MB modules. That, and IBM changes their FRU's like most people change their socks. Found several I think will work but need to cross reference the FRU's.
Guess that's the way it goes! :)